| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* INRA, UR1213 Herbivores, Site de Theix, F-63122 Saint-Genès-Champanelle, France
INRA, UMR791 Physiologie de la Nutrition et Alimentation, INAPG, 16 rue Claude Bernard, F-75231 Paris, France
1 Corresponding author: fglasser{at}clermont.inra.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: dairy cow • duodenal flow • milk fatty acid • desaturation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Due to the extensive biohydrogenation of unsaturated FA in the rumen (Doreau and Ferlay, 1994; Glasser et al., 2008), FA intakes are not a representative indicator of the FA actually available for the animal. For this reason, we only used publications reporting FA flows at the duodenum and postruminal infusions of FA, which are more representative of FA availability for milk fat synthesis. The aim of the present study is to quantitatively determine, from the published data, the response of milk FA yield and C18 composition to duodenal C18 flow.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Inclusion of Publications
We compiled all available experiments on dairy cows (published until 2006) reporting duodenal FA flows, milk fat yield and milk FA profiles, or any set of data that could be used to calculate these criteria. From this data set, we excluded 1 experimental treatment with saturated tallow and very low intestinal digestibility (Pantoja et al., 1996), 1 experiment containing discrepancies between FA intake and duodenal flows (Loor et al., 2002), and 1 experiment comparing different forages with almost no difference between treatments in duodenal C18 flow (Dewhurst et al., 2003), thus not relevant for our meta-analysis. The final data set of dietary experiments included 16 experiments reported in 21 publications (Klusmeyer and Clark, 1991; Tice et al., 1994; Wonsil et al., 1994; Pantoja et al., 1996; Enjalbert et al., 1997; Kalscheur et al., 1997a,b; Christensen et al., 1998; Avila et al., 2000; Piperova et al., 2002; Shingfield et al., 2003; Ueda et al., 2003; Chelikani et al., 2004; Gonthier et al., 2004; Loor et al., 2004, 2005b, c,d; Lundy et al., 2004; Qiu et al., 2004; Gonthier et al., 2005). To this data set, we added all available experiments on duodenal infusion of unsaturated fats reporting milk fat yield and FA composition (i.e., 13 experiments reported in 12 publications; Chilliard et al., 1991a,b; Christensen et al., 1994; Gaynor et al., 1994; Ottou et al., 1995; Bandara, 1997; Enjalbert et al., 1998; Wagner et al., 1998; Enjalbert et al., 2000; Romo et al., 2000; DePeters et al., 2001; Bell and Kennelly, 2003). Finally, to test the possible impact of de novo synthesis on C18 FA yield, independently to duodenal C18 flow, we also included experiments on postruminal infusions of trans-10, cis-12–18:2, which is a potent inhibitor of de novo synthesis (Baumgard et al., 2000). Infusions with a significant difference (P < 0.05) in DMI among experimental treatments were excluded to ensure that duodenal C18 flows could be considered equal between control and infused treatments. Eleven studies on trans-10, cis-12–18:2 infusions were thus included (Chouinard et al., 1999a,b; Baumgard et al., 2000, 2001, 2002; Bell and Kennelly, 2003; Loor and Herbein, 2003; Mackle et al., 2003; de Veth et al., 2004; Perfield et al., 2004; Saebo et al., 2005). In total, our final database included 39 experiments and 139 experimental treatments. All but 2 experiments were conducted on Holstein or Friesian cows. The mean forage percentage was 52% (on a DM basis, range 35 to 100%), the main forage being alfalfa silage or haylage in 11 experiments, corn silage in 10 experiments, alfalfa hay in 9 experiments, grass silage in 6 experiments, grass hay in 2 experiments, and fresh grass in 1 experiment. Three of the experiments were conducted on early lactation cows (<30 DIM), 24 on midlactation cows (30 to 150 DIM), and 12 on late-lactation cows (>150 DIM).
Several equations were derived from this database. For each equation, a subset of experiments was selected based on their relevance for the studied relationship. The selection criteria, number of experiments, and experimental treatments used for adjustment of each equation are described under their corresponding paragraph headings.
Calculation of Flows and Mammary Desaturation
Total milk FA yields were computed from the milk fat yields reported in the publications, assuming that total FA represent 93.3% of milk fat (Glasser et al., 2007a). Although some publications directly provided FA yield data, we recalculated the data from milk fat yields according to the same method to ensure homogeneity among publications. The same method was used to calculate the amounts of duodenal FA provided by oil infusions (considered as 100% triglycerides, containing 95% FA).
We chose to evaluate mammary de novo FA synthesis by milk C4 to C16 yield. Although part of C16 (and probably, to a lesser extent, C14 as well) is derived from mammary uptake, the yield of C16 was highly correlated, within experiment, to the sum of C4 to C14 yields (P < 0.001), but not to C18 yield (P = 0.20). The yield of C16 was thus probably for a large part under the same regulation as the short- and medium-chain FA, and consequently, we chose to consider the sum C4 to C16 as a whole. We only summed the straight- and even-chain FA of C4 to C16 (both saturated and monounsaturated), because odd- and branched-chain FA were rarely measured in our data set and they always represented relatively small amounts compared with the even-chain FA. In 4 publications not providing complete C4 to C16 data, we estimated this amount as 0.944 x (total FA – C18 to C22 FA) (the 0.944 coefficient was obtained from the remaining studies that had complete data).
To estimate the mammary desaturation of 18:0, trans-11– and trans-13–18:1 (i.e., the main C18 substrates of 
9-desaturase), we had to estimate the mammary uptake of the products and substrates of desaturase. We proceeded in 3 steps: i) we estimated the transfer efficiency of the sum (substrate + product) between duodenum and milk (proportion of the duodenal flow that is secreted in milk); ii) we then applied this common transfer efficiency separately to the duodenal flows of the substrate and of the product, thus producing an estimate of the mammary uptake of both substrate and product [this hypothesis of a similar transfer efficiency of substrate and product is supported by very similar mammary extraction rates (estimated from arteriovenous differences) for 18:0 and cis-9–18:1 (Bickerstaffe et al., 1972; Enjalbert et al., 1998; Loor and Herbein, 2003) and for trans-11–18:1 and cis-9, trans-11–18:2 (Loor and Herbein, 2003)]; and iii) we then took the milk yield of the product minus its estimated mammary uptake and estimated the mammary desaturation.
Statistical Analyses
We analyzed the data with GLM models (Minitab, version 13). An experiment effect (fixed effect) was introduced in each model, meaning that the resulting models are within-experiment models (reflecting the biological effects of the independent variables, e.g., duodenal C18 flow). The experiment effects (noted exp in the equations, including various components, e.g., animal genetics, flow and FA measurement methods, etc.) do not thus interfere with the resulting adjustments. When quadratic coefficients were found to be significant, we used Akaike’s information criterion to choose the best fit between linear and quadratic models. No significant effect of lactation stage was found in any of the models tested (but the data were not balanced for this effect). For each model, the number of experiments (Nexp) and experimental treatments (Ntrt) used to adjust the model are indicated, as well as R2 and the root mean square error (RMSE).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), but with wide variation in slopes and also a large within-experiment variation in milk C18 yield for the same duodenal flow. The low-lipid diets of all experiments, defined as the diets containing less than 30 g of FA/kg of DMI or 35 g of fat/kg of DMI (based on data distribution, white circles in Figure 1) were aligned along a line (dashed line in Figure 1, slope = 0.85). For these low-lipid diets, milk C18 yield was also proportional to DMI and milk yield.
|
. Based on graphic examination and distribution of within-experiment differences, we identified 2 conditions in which milk C18 yield did not respond to duodenal C18 flow. The first condition was observed in treatments with similar duodenal flow (less than 20 g/d of difference between treatments in a same experiment) and yet differences in milk C18 yield (greater than 60 g/d), corresponding to vertical lines in Figure 2 (Christensen et al., 1994; Gaynor et al., 1994; Enjalbert et al., 1998; Romo et al., 2000). The second condition was observed in treatments with an increase in duodenal C18 flow (more than 80 g/d of difference between treatments in a same experiment) and yet almost no increase (less than 10 g/d) or even a decrease in milk C18 yield, corresponding to horizontal or decreasing lines in Figure 2 (Chilliard et al., 1991b; Kalscheur et al., 1997a,b; Chelikani et al., 2004; Loor et al., 2005b,c). These treatments are indicated by black triangles in Figure 2. For these treatments, C18 yield was obviously not determined by C18 availability. These unresponsive treatments were excluded from the global model of the response of milk C18 yield to an increase in C18 availability (increase in duodenal C18 flow) and considered apart (see below). Moreover, we excluded diets containing fish oil (white triangles in Figure 1
), because they induced a large decrease in C18 yield for a given duodenal flow (Figure 1). The fish oil treatments were thus studied apart (see fish oil effects paragraph below).
|
C18duo):
 | [1] |
The intercepts and linear and quadratic coefficients of the equation did not differ between dietary and infusion experiments. There was a linear within-experiment decrease in DMI with increasing duodenal C18 flow [mean slope –1.7 (±0.5) g of DMI per g of duodenal C18 increase; P < 0.001], but variations in DMI and milk C4 to C16 yield, when introduced as covariates in the Eq. [1] model, were not significant.
To study the determinants of milk C18 yield in the unresponsive treatments, we plotted all data according to their milk yields of C18 and C4 to C16 (Figure 3). Although most experimental treatments exhibited the well-known negative relationship between milk C18 and C4 to C16 yields, the unresponsive treatments were aligned along a line (dashed line in Figure 3), indicating a positive relationship between C18 and C4 to C16 yields, in all but 1 case (18:0 infusion of Enjalbert et al., 1998). The within-experiment adjustment of this line was:
|
 | [2] |
The intercept was not significant (P = 0.23). These treatments have a mean percentage of 45.2 ± 1.5% C4 to C16 and 51.7 ± 1.3% C18 in total milk FA. In these treatments, C18 yield was thus proportional, within experiment, to C4 to C16 yield, but neither to milk yield (P = 0.90) nor to DMI (P = 0.69). These treatments are referred to hereafter as highly loaded in C18. The only treatment not aligned with the others was a duodenal infusion of stearic acid (Enjalbert et al., 1998), in which C18 yield must have been limited by other processes.
Moreover, in Figure 3, most of the low-lipid diets (white circles) were also situated around a line (solid line in Figure 3, Y = 0.55X), reflecting a positive relationship between milk C18 and C4 to C16 yields, mainly driven by differences in milk yield of cows between experiments. The mean composition of total milk FA in these low-lipid diets was 61.3 ± 1.5% C4 to C16 and 34.8 ± 1.2% C18. In most experiments, lipid supplementation of the low-lipid control diets induced both a decrease in C4 to C16 yield and an increase in C18 yield, corresponding to the well-known inverse relationship between the 2 FA groups, represented by a shift in the diagonal toward the upper left of Figure 3 after lipid supplementation.
Effect of De Novo Synthesis on C18 Yield
In the unresponsive treatments, we observed a positive relationship between C4 to C16 and C18 yields (see above, Eq. [2]), suggesting a possible relationship between these 2 FA groups. However, in most experiments, there were simultaneous variations in duodenal C18 flow, DMI, and milk C4 to 16 and C18 yields. To study whether there was a specific effect of de novo synthesis (milk C4 to C16 yield) on milk C18 yield independently of duodenal C18 availability, we selected the experiments reporting a decrease in milk C4 to C16 yield with a constant C18 availability. To this end, we selected experiments (or subsets of experimental treatments within experiments) with an intraexperiment variation in C4 to C16 yield and a supposedly constant duodenal flow of C18 (no significant variation in DMI, or a similar duodenal flow or a similar amount of C18 infused). These experiments were extracted from 3 sets of experiments: trans-10, cis-12–18:2 infusions, duodenal infusions of other lipids (e.g., comparison of different lipid sources in similar amounts), and dietary experiments. In trans-10, cis-12–18:2 infusions, milk C4 to C16 yield was found to decrease in a quadratic manner according to the amount of trans-10, cis-12–18:2 infused (data not shown).
As seen in Figure 4, almost all these experiments exhibited positive slopes (i.e., a decrease in milk C18 yield when milk C4 to C16 yield decreased), regardless of whether they were trans-10, cis-12–18:2 infusions (triangles) or other lipid infusions or dietary experiments (circles). However, the slopes differed between experiments. To find a global relationship including all these experiments, we tested the relationship between the slope of the response of C18 yield to C4 to C16 yield and the distance to the high C18 load line. For each experiment in Figure 4 (n = 21), we computed the mean slope and the distance of the midpoint of the experiment to the high C18 load line (dashed line, corresponding to Eq. [2], adjusted from data in Figure 3). There was a negative linear relationship between within-experiment slope and distance to the line:
|
 | [3] |
Both the intercept and the slope of this relationship were significant (P < 0.01). Slopes were inferior to 1, meaning that C18 yield decreased less than C4 to C16 yield, and there was a (linear) increase in C18 percentage in milk FA after a decrease in milk C4 to C16. The within-experiment slope obtained for a null distance to the line (0.78, i.e., when the experiment is very close to the high C18 load line) is close to the slope of the line itself (0.92, Eq. [2]), obtained from a different set of data. The slopes did not differ between dietary and infusion experiments. For the experiments with a low percentage of C18 in milk FA (<45%, white symbols in Figure 4), the mean within-experiment slope was 0.30 (±0.09).
Effect of Duodenal Flow of C18 on C4 to C16 Yield
It is well known that lipid supplementation decreases de novo FA synthesis, particularly through certain rumen biohydrogenation isomers (Bauman and Griinari, 2001). There were few publications in the database with sufficiently detailed profiles of C18 at the duodenum. Moreover, there was often a high correlation between the duodenal flows of the various C18 isomers, because plant oil supplements (which were the most widely used in the database) induce higher duodenal flows of almost all C18 isomers (mainly intermediates of rumen biohydrogenation of polyunsaturated FA). We compared several of these isomers as predictors of a decrease in C4 to C16 yield, but none proved better (based on the RMSE of the models) than the duodenal flow of total C18 (Figure 5). Within-experiment in dietary and lipid (except trans-10, cis-12–18:2) infusions, milk yield of C4 to C16 decreased linearly with increasing C18 duodenal flow:
|
 | [4] |
There was a significant effect of infusions vs. dietary experiments on the intercept (the intercept value in Eq. [4] is the overall mean), but not on the slope. There was no significant effect of DMI variation added as a covariate in this model (P = 0.30).
Fish Oil Effects
There were 4 experiments (6 experimental treatments) in the database reporting fish oil supplementation (Wonsil et al., 1994; Shingfield et al., 2003; Qiu et al., 2004; Loor et al., 2005c). The fish oil intakes varied between 250 and 430 g/d, with duodenal flows of C20 to C24 FA ranging from 9 to 18 g/d. In each experiment, we estimated the milk C18 and C4 to C16 yields of the fish oil treatments based on the respective treatment(s) without fish oil and Eq. [1] and Eq. [4] (models adjusted from diets without fish oil). We then compared these estimated yields (that would be obtained without fish oil by an equivalent amount of duodenal C18) with the data from the fish oil treatments reported in the publications. For both milk C18 and C4 to C16, the recorded yields for fish oil diets were significantly lower than the yields predicted by Eq. [1] and Eq. [4] from their duodenal C18 flows. Recorded data were 38% (range 20 to 45%) lower than the modeled data for milk C18 yield (equivalent to a decrease of 100 to 164 g/d) and 17% (8 to 29%) lower for milk C4 to C16 yield (equivalent to a decrease of 30 to 175 g/d). Most fish oil treatments proved very similar to the low-lipid diets in terms of C4 to C16 and C18 proportions, with a low percentage of C18 in milk FA (mean 36%).
Milk C18 Composition
The wide variability in milk C18 yields for similar duodenal flows of total C18 (see above and Figure 1) precludes the prediction of the yield of a specific milk C18 FA based on its duodenal flow, particularly for high-lipid diets. We thus chose to study duodenal and milk C18 from the composition point of view [and not with a flow approach as used above for total C18]; i.e., to compare the percentages of individual C18 in total milk C18 to their respective percentages in total duodenal C18. To take into account the existence of mammary 9-desaturation activity, we summed the products and substrates of 9-desaturase in duodenal and milk C18 contents (i.e., the sum of 18:0 and cis-9–18:1, the sum of trans-11–18:1 and cis-9, trans-11–18:2, and the sum of trans-13–18:1 and cis-9, trans-13–18:2). The data of duodenal and milk percentages of the sum (trans-11–18:1 + cis-9, trans-11–18:2) are in Figure 6a. For 18:3, we excluded certain publications reporting higher or much lower milk secretions than duodenal flows (Klusmeyer and Clark, 1991; Tice et al., 1994; Pantoja et al., 1996; Avila et al., 2000; Shingfield et al., 2003; Qiu et al., 2004). Table 1 presents the adjusted models of milk C18 percentages according to duodenal C18 [the model for the sum (trans-11–18:1 + cis-9, trans-11–18:2) is also in Figure 6b]. Except for 18:3, all regressions gave an R2 of over 0.79, meaning that the percentages of FA in milk were closely related to their percentages at the duodenum. Five models had slopes significantly lower than 1 (4 with 0.01 < P < 0.05, only trans-9–18:1 with P < 0.01), but all the slopes were numerically lower than 1, and all intercepts were positive, except those of the 18:3 model. From the adjusted equations, it was possible to compute, for each FA, the value for which percentage in milk equals percentage in duodenum (intersection of the model equation with the first bisector, see Figure 6b). These values are very close of the mean C18 composition observed at the duodenum (Table 1), except for 18:3, for which percentage in milk C18 is lower than percentage in duodenal C18 over all the variation range. This means that for all FA except 18:3, when they are at a rather low level at duodenum (lower than the mean values), their level in milk tends to be greater.
|
|
9-desaturation substrates studied (18:0, trans-11– and trans-13–18:1), we adjusted the models of the amount desaturated (endogenous mammary production) according to estimated mammary uptake. Figure 7 presents a plot of the estimated endogenous production of cis-9–18:1 (desaturation product) according to the estimated mammary uptake of 18:0 (desaturation substrate). The 3 resulting within-experiment models are given in Table 2. All models were linear, all the intercepts were not significantly different from 0, and all the slopes were highly significant (P < 0.001), meaning that, within experiment, desaturation was directly proportional to the mammary uptake of the substrate. We also estimated the contribution of mammary desaturation to the yield of the desaturation product. For cis-9–18:1, the canola oil infusion reported by Chelikani et al. (2004) was included in the data used for model adjustment, and it fitted well with the other data but resulted in a very low contribution of mammary desaturation of 18:0 in the milk yield of cis-9–18:1 (15.8%), probably due to the high cis-9–18:1 supply from canola oil infusion. For the other treatments, the mean contribution was 80.2 ± 2.3% (range: 48.6 to 94.0%). For cis-9, trans-11–18:2, the estimated contribution of mammary desaturation to the milk yield was extremely low (3.6%) in the control treatment of Shingfield et al. (2003), with duodenal flow and milk yield almost equal for both trans-11–18:1 and cis-9, trans-11–18:2. For the other data, the mean contribution of mammary desaturated trans-11–18:1 to milk cis-9, trans-11–18:2 yield was 94.7% ± 1.8% (range: 71.6 to 100%). The mean contribution of mammary desaturated trans-13–18:1 to milk cis-9, trans-13–18:2 yield was 81.1% ± 4.4% (range: 58.6 to 100%).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. However, different configurations can be distinguished with respect to their relationship.
First, the study of cows fed low-lipid diets in the different experiments mainly reflected the effect of the production level of the cows varying between experiments, illustrated by positive correlations between DMI, milk yield, duodenal C18 flows, and milk C4 to C16 and C18 yields: milk yields varied between 11.4 and 41.7 kg/d, with a mean fat content at 34.1 g/kg and 34.8% of C18 in milk FA. With these diets, milk C18 yield represented a large proportion (around 85%, slope of the line in Figure 1) of duodenal C18 flow. We cannot exclude that C18 from lipomobilization contributed to milk C18 secretion, even if most of the data were obtained in mid- to late-lactation cows, for which this lipomobilization-based contribution is probably low (Chilliard et al., 1991b).
Second, when data were studied with a within-experiment approach, in most cases, an increase in duodenal C18 flow increased milk C18 yield (Figure 2) and in a quadratic fashion (Eq. [1]). The transfer efficiency of the supplemental C18 was relatively low (below the linear coefficient of 0.46). The resulting milk C18 yield was maximal for an increase in duodenal flow of 625 g/d and decreased thereafter.
Finally, for some high-lipid diets, milk C18 yield either varied greatly despite similar duodenal flows or did not respond to duodenal flow increase (black triangles in Figure 2). In these cases, as seen in Figure 3, milk C18 yield was proportional to milk C4 to C16 yield (Eq. [2]), meaning that the variation in de novo synthesis in these high-lipid diets could explain the milk C18 yield response. These milks were characterized by a high percentage of C18 (51.7%) in total FA. This decrease in milk FA yields was not related to a decrease in milk yield, which was only slightly affected by most lipid supplements, even in severe milk fat depression (Mackle et al., 2003).
This hypothesis whereby de novo synthesis may to a certain extent drive milk C18 yield was confirmed by the study of the experiments displaying a variation in milk C4 to C16 yield in a context of constant C18 availability (pooling dietary experiments, experiments involving duodenal infusion of unsaturated lipids, and experiments involving trans-10, cis-12–18:2 infusions). In these experiments, the treatments decreasing milk C4 to C16 yield also decreased C18 yield (Figure 4), despite a similar availability from the intestinal tract. A similar pattern was observed in the individual responses of cows to lipid supplementation from 2 experiments included in the present meta-analysis (Glasser et al., 2007b), in which a low-forage diet, inducing a drop in de novo FA synthesis, restrained the incorporation of C18 from lipid supplement in milk fat.
The explanation of this effect may lay at the mammary FA esterification step, which involves both de novo synthesized FA and long-chain FA taken up from plasma. Mammary FA esterification involves several enzymes that successively esterify FA on the 3 carbons of the glycerol backbone to produce triglycerides (TG), which account for 95 to 98% of milk lipids (Jensen, 2002). The enzyme responsible for the last esterification (i.e., on carbon sn-3 of the glycerol) is the diacylglycerol acyl transferase, and this sn-3 position is characterized by a high percentage of short-chain FA, whereas the opposite is observed with the sn-1 position, which is esterified with a large majority of long-chain FA (C16 and C18). This esterification hypothesis could explain at least part of the relations observed between the 2 groups of FA (C4 to C16 and C18). In a context of high duodenal C18 flow, the limiting factor would probably be the availability of short-chain FA for the last step of esterification, and C18 incorporation in milk lipids could be limited by de novo synthesis of short-chain FA, leading to the concept of a milk fat highly loaded in C18. Even if we cannot completely rule out a direct effect of the treatments (particularly trans-10, cis-12–18:2 infusions) on C18 mammary uptake itself (rather than an effect through de novo synthesis inhibition), 2 observations undermine this hypothesis: i) this phenomenon was also observed in dietary experiments and experiments involving plant lipid duodenal infusion (i.e., without increasing trans FA from the rumen), and ii) among trans-10, cis-12–18:2 infusions, the slope of the decrease in C18 yield to the decrease in C4 to C16 yield varied (Figure 4 and Eq. [3]), which would not be the case if both yields were regulated independently by trans-10, cis-12–18:2.
Unsaturated lipid supplementation often induces a decrease in mammary de novo synthesis (Chilliard et al., 2000). Based on the database used here, milk C4 to C16 yield was found to decrease linearly with increasing duodenal C18 flow (Eq. [4]). The inhibiting effect of C18 on de novo synthesis has been known for some time, but the precise mechanisms involved are not yet well understood, even if some 18:2 isomers, such as trans-10, cis-12–; cis-10, trans-12–; trans-9, trans-11–; or trans-9, cis-11–18:2 have been identified or suspected as being inhibitors (Baumgard et al., 2000; Bauman and Griinari, 2001; Saebo et al., 2005; Roy et al., 2006; Shingfield and Griinari, 2007). Because lipid supplementation leads to a simultaneous increase in the duodenal flows and milk concentration of many isomers, as well as a decrease in DMI (Allen, 2000), the origin of the observed decrease in C4 to C16 yield is impossible to allocate between the inhibition by one or several of these isomers, a decrease in DMI or a substrate competition of C18 on short-chain FA esterification. The available experiments did not enable to disentangle these effects. Another study found the substrate availability for de novo synthesis (evaluated by NDF intake) and duodenal conjugated linoleic acid flow to have an additive effect on milk C4 to C16 yield (Glasser et al., 2007b). The model of a linear decrease of milk C4 to C16 with increasing duodenal total C18 flow is the result of all these effects (within the data set used), and has the advantage of simplicity, but would probably deserve to be improved (from a larger data set), because it does not take into account that the forage content and nature of the basal diet influence milk C4 to C16 yield and response to lipid supplements in cows (Chilliard et al., 2001, 2007; Dewhurst et al., 2006), probably through modifications of rumen FA metabolism and production of precursors of de novo synthesis.
An increase in duodenal C18 flow after dietary lipid supplementation would thus induce a linear decrease in C4 to C16 yield and a quadratic increase in milk C18 yield. The relatively low increase in milk C18 yield and decrease over 625 g/d of supplemental duodenal C18 (Eq. [1]), under this hypothesis, could be partly explained by the decrease in C4 to C16 yield. Indeed, when lipid supplementation does not induce a decrease in de novo synthesis (e.g., protected lipid sources), there are high increases in milk C18 yield (Bartsch et al., 1976; Goering et al., 1977; Mansbridge et al., 1999; Goodridge et al., 2001).
Increasing duodenal C18 flow thus leads to an increase in C18 percentage. It could be posited that, above a certain threshold, milk lipids become highly loaded in C18 and that a further decrease in de novo synthesis induces a rapid decrease in milk C18 incorporation, as seen in Figure 3 and as described by Eq. [2]. This could explain some very low transfer efficiencies of C18 FA between duodenum and milk when C4 to C16 yield was low (e.g., the low forage + linseed oil diet in Loor et al., 2005b) and the absence of a response to oil infusions in early lactation cows presenting a high percentage of C18 in milk FA even without oil infusion (Chilliard et al., 1991b). According to our data, this threshold would be situated around 51.7% but may depend on other factors, including C18 composition. This percentage is very close to the percentages observed in early lactation cows receiving either a low-lipid diet or a duodenal infusion of rapeseed oil (51 to 54% in Chilliard et al., 1991b), suggesting that both lipid mobilization and intestinal C18 supply could highly load milk TG in C18. The closer the experiment to the high C18 load line, the more the C18 yield would decrease in response to C4 to C16 inhibition, as seen in the relation between the slope of the relationship and the distance to the high C18 load line (Figure 4). When C4 to C16 yield was inhibited in cows that had milk fat already highly loaded in C18 (as in trans-10, cis-12–18:2 infusion in lipid-supplemented diets; Loor and Herbein, 2003), milk C18 yield decreased along the high C18 load line, and this decrease was around 83 to 85% of the decrease in C4 to C16 yield, thus leading to a further slight increase in the percentage of C18 in total milk FA.
At the opposite, in cows fed low-lipid diets, the low C18 availability may limit the incorporation of short-and medium-chain FA in milk TG to maintain a minimum percentage of C18 at around 35% of total milk FA. The production of diacylglycerols (mainly composed of long-chain FA), which are a substrate of diacylglycerol acyl transferase, would remain low and would thus limit the incorporation of short- and medium-chain FA in milk TG. In this context, low levels of C18 supplementation could act as primers of TG synthesis and increase both de novo synthesis and C18 incorporation in milk lipids.
Our hypotheses are consistent with the results of a study published after the constitution of our database, comparing increasing levels of abomasal lipid infusion (Drackley et al., 2007). The data calculated from this study, together with the within-experiment models Eq. [1] and Eq. [2], are in Figure 8. Milk C18 yield increases quadratically, according to Eq. [1], from 0 to 500 g/d of FA infusion (corresponding to 51.8% of C18 in total milk FA, very close to the mean of 51.7% found from our database), and then decreased linearly, close to the Eq. [2] model. Milk C18 yield with 1,000 g/d of FA infusion was even lower than the control one. Even if the models would need a validation on a larger data set (unfortunately, experiments of increasing amounts of lipid supplementation are very scarce), the good agreement with this independent study is promising.
|
Finally, we have to underline the fact that a meta-analysis depends on the available data set (the meta-design). In the available publications reporting duodenal FA flows and milk FA composition, there was almost always a covariation between DMI and duodenal C18 flow (with a mean slope of –1.7 g/g), no experiment being available, to our knowledge, which was designed to compare separately level of intake and lipid supplementation. In all models, we tested the possible contribution of DMI added as a covariable, which was never significant. We had thus to assume that the adjusted coefficients took into account the effect of the covariation in DMI. They are thus probably somewhat biased but are the best estimates in the current state of published knowledge and usable provided that the decrease in DMI associated with duodenal C18 flow increase is not too far from the mean –1.7 g/g observed in the present data set.
Milk C18 Composition
The comparative study of duodenal and milk C18 profiles (individual C18 expressed as a percentage of total C18) revealed that most FA percentages in milk were closely related to their percentages in duodenal C18, when the products and substrates of 9-desaturase were summed. Moreover, all the equations but one were close to the bisector, meaning that the C18 profiles in milk were close to those at the duodenum. For an average duodenal composition, almost all C18 FA thus seemed to be similarly transferred from duodenum to milk. For duodenal C18 composition diverging from the average values, there could be some buffering capacity of the udder, the FA in low percentage seeming positively selected (percentage in milk slightly higher than at duodenum) and reciprocal for the FA in high percentage. The relationships obtained for cis-9, cis-12–18:2 (low slope) and 18:3 (low R2 compared with the others) may be explained by the metabolism of these essential FA, which are highly incorporated in plasma polar lipids.
The similarities and differences between duodenal and milk profiles may stem from intestinal absorption or mammary uptake of the FA or both. Concerning FA absorption from the intestine, it is known that there are some differences in apparent intestinal digestibilities of the various C18 (Doreau and Ferlay, 1994; Glasser et al., 2008), which could cause some differences between absorbed and duodenal profiles of C18. However, given the relatively scarce published data of apparently absorbed flows, it was not possible to adjust the milk C18 percentages from absorbed C18 flows. Data concerning mammary uptake of FA from plasma TG are inconsistent: from labeled FA infusions or comparisons between pre- and postmammary TG composition, some authors have reported a similar uptake of the different FA (Bickerstaffe et al., 1972), whereas others have reported that the uptakes are different (Annison, 1967; Thompson and Christie, 1991). Based on our approach, we could not detect any major differential transfer rate between C18 FA (except, to a certain extent, for cis-9, cis-12–18:2 and 18:3), but we cannot rule out a compensation effect between differences in intestinal absorption and differences in mammary uptake.
FA Desaturation
The estimation of cis-9–18:1; cis-9, trans-11–; and cis-9, trans-13–18:2 syntheses by mammary desaturation of 18:0, trans-11–, and trans-13–18:1, respectively, showed that the desaturated amounts were directly proportional to the mammary uptake of the precursors, with little variation among experiments. Desaturation capacity did not seem to be limiting within the range of precursor uptakes estimated in the data set (18:0 uptake up to 400 g/d, trans-11–18:1 uptake up to 70 g/ d, and trans-13–18:1 uptake up to 26 g/d). Our estimation of mammary uptake did not take into account circulating FA from the lipolysis of body lipid reserves, because most experiments were conducted on mid- to late-lactation cows, which are in positive energy balance without net lipid mobilization. However, there is always a body lipid turnover (Chilliard, 1993), and we cannot exclude, even in cows in positive energy balance, a contribution of FA from lipomobilization to milk FA, especially for 18:0 and cis-9–18:1 (Chilliard et al., 1991b). However, the nonsignificant intercepts of the relationships exclude a major contribution of FA from adipose tissue turnover (that are more unsaturated than duodenal FA). The slope of the cis-9–18:1 model (Table 2) was 0.54, meaning that around 54% of the additional 18:0 uptake in lipid-supplemented diets was desaturated. This estimate is consistent with estimates obtained by labeled FA or mammary balances, at 47 to 55% (Lauryssens et al., 1961; Bickerstaffe and Annison, 1974; Enjalbert et al., 1998; Mosley and McGuire, 2007), although other authors concluded lower estimates at between 25 and 45% (Annison, 1967; Linzell et al., 1967; Annison et al., 1974).
The mean contributions of mammary desaturation to milk yield of cis-9–18:1 and cis-9, trans-13–18:2 were estimated at around 80%. For cis-9–18:1, this figure was similar to the results obtained from mammary uptake studies in goats [i.e., 80 to 84% (Annison et al., 1974)] but higher than in cows, where it varied from 45 to 70% (Bickerstaffe and Annison, 1974; Enjalbert et al., 1998; Mosley and McGuire, 2007). The lowest value estimated in the present study (15.8%), for an abomasal infusion of a canola oil (rich in cis-9–18:1; Chelikani et al., 2004), was close to that obtained from mammary uptake measurements during cis-9–18:1 duodenal infusion (11%, Enjalbert et al., 1998). For cis-9, trans-11–18:2 mammary desaturation as a function of trans-11–18:1 mammary uptake, averages highly differed between experiments (perhaps due to analytical techniques), inducing a wide variability in the absolute desaturation ratios, but when the experiment effect was taken into account, the within-experiment slopes remained very homogenous among experiments. The mean slope was 0.21, meaning that 21% of the additional trans-11–18:1 uptake in lipid-supplemented diets was desaturated in cis-9, trans-11–18:2 in the udder (in line with the 25.7% estimate from labeled trans-11–18:1 in Mosley et al., 2006 or 28.9% from an infusion experiment in Shingfield et al., 2007). This figure is very close to the figure obtained for trans-13–18:1 desaturation (0.22), contradicting in vitro data on rat microsomes indicating a higher desaturation of trans-13–18:1 compared with trans-11–18:1 (Mahfouz et al., 1980). Overall, the results in cows suggest that 18:0 is proportionally more desaturated by mammary cells than trans–18:1 isomers.
The mean contribution of mammary desaturation to milk cis-9, trans-11–18:2 yield was estimated at 94.7%. This figure is in line with but somewhat higher than estimates obtained with sterculic acid (a potent inhibitor of desaturation) infusions (64% in Griinari et al., 2000; 78% in Corl et al., 2001; more than 87% in Kay et al., 2004), in dietary experiments (82 to 87% in Lock and Garnsworthy, 2002; more than 93% in Piperova et al., 2002; 82 to 97% in Loor et al., 2005c), and with labeled trans-11–18:1 (83% in Mosley et al., 2006). This estimation is highly dependent on the measured duodenal flow of cis-9, trans-11–18:2, as shown by the extremely low value (4%) estimated from Shingfield et al. (2003), who measured a high cis-9, trans-11–18:2 duodenal flow.
In conclusion, despite the low number of experiments reporting both duodenal flow and milk yield of FA, the meta-analysis of these data and duodenal infusion experiments made it possible to build hypotheses regarding the regulation of cow mammary lipogenesis. The present study underlines the interdependence of C18 and C4 to C16 yields. This regulation most probably lies at the esterification step of milk fat synthesis, which involves both long-chain and de novo synthesized FA. In low-lipid diets, milk C18 yield is probably limited by C18 supply. Milk fat yield in response to lipid supplementation (inducing an increase in duodenal C18 flow) is the result of two additive phenomena: an induced quadratic increase in milk C18 yield and an induced linear decrease in C4 to C16 yield. A high decrease in de novo synthesis beyond a certain threshold, which would correspond to milk fat highly loaded in C18, could dramatically limit, or even almost totally prevent, any increase in milk C18 yield, whatever its availability. The transfer efficiency of duodenal C18 to milk is thus highly variable. However, despite this variability, the milk C18 profiles are very close to the respective profiles at the duodenum, once mammary desaturation is taken into account. Moreover, the mammary desaturation of 18:0, trans-11–, and trans-13–18:1 is proportional to their mammary uptake and contributes 80 to 95% to the milk yield of their 9-desaturated products. These models and hypotheses could be challenged by a wider set of experiments on dietary lipid supplementation reporting not duodenal flow data but FA intake, as far as C18 duodenal flows could be estimated from FA intakes.
Received for publication May 24, 2007. Accepted for publication March 13, 2008.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
9-desaturase in the production of cis-9, trans-11 CLA. J. Nutr. Biochem. 12:622–630.[CrossRef][Medline]9-desaturase. J. Nutr. 130:2285–2291.9-desaturation of myristic, palmitic, and stearic acids in lactating dairy cattle. Lipids 42:939–945.[CrossRef][Medline]This article has been cited by other articles:
|
|
 |
|
  F. Glasser, A. Ferlay, and Y. Chilliard Oilseed Lipid Supplements and Fatty Acid Composition of Cow Milk: A Meta-Analysis J Dairy Sci, December 1, 2008; 91(12): 4687 - 4703. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
